Self Driving RC Car/2020

Videos

Open Source

Presentation

Source Code

Parts List

Assembly

Raspberry

PI Installation

Server Installation

Deployment

First Experiments

[edit]

[edit]

OsProject
edit
id	dukes
state
owner	rc-dukes
title	Dukes of Hazzard self driving RC car
url	https://github.com/rc-dukes/dukes
version	0.0.2
description
date	2019-07-08
since
until

[edit]

[edit]

• https://github.com/rc-dukes/dukes

In the javadoc apidocs you'll also find generated uml-diagrams like this one: https://rc-dukes.github.io/dukes/dukes/apidocs/nl/vaneijndhoven/daisy/package.svg

[edit]

Parts-List as Excel File

#	picture	part	example sources	documents	~ price
1		1:10 RC Car chassis	1:10 elektro tourenwagen chassis ARR TAMIYA 57986" 1:10 RC TT-02-Chassis First Try On-Road	de/en/fr/nl Tamiya 57986	70 €
2		Speed Control	Modelcraft Carbon Series Speed Control "20 Turns"	de/en/fr/nl	21 €
3		RC-Car servo	RC-Car servo 4519 Carson 500502036 - Servo CS-6 - 6 kg/JR-Stecker	de/en	10 €
4			TopFuel LiPo 25C-ECO-X 1300mAh 3S  NiMH 7.2V 3500 mAh		16 €
5		Battery Charger	Ansmann NiMH Charger		27 €
6		Raspberry PI 3 B+	Raspberry PI 3 B+	en	35 €
7		microSD Card	SanDisk 16 GB		7 €
8		Raspberry PI Camera	1080 p Camera Module		23 €
9		DC/DC Converter 12 V-> 5V	12 V->5 V Converter Module		7 €
10		Micro USB Connector	Delock USB 2.0 Micro USB		6 €
				Total	199 €
a		Breadbord Kit	MB102 Breadbord Kit		7 €
b		Ulrasound Sensor	5 x HC-SR04 + Cables		10 €

[edit]

DCC-DCC converter

To run the Raspberry PI from the 7.4 Volt Battery of the car the DC-DC converter has to be soldered into the voltage supply chain. You might want to select a converter with enough power for your usecase

15 Watt 3 A

This configuration works if you only want to run a Raspberry PI and directly control the motor and steering servos of the RC car.

• 
• 
• 
• 

25 Watt 5 A

This configuration is currently tried for running the servos over an Adafruit 16 channel servo controller and using extra servos.

Camera mount

• 
• 

Servo and LED

[edit]

For a start a 3D Printed case for the Raspberry PI was used:

[edit]

see Raspberry PI

WebCam Interface

• https://github.com/silvanmelchior/RPi_Cam_Web_Interface

cd /usr/local/src
sudo git clone https://github.com/silvanmelchior/RPi_Cam_Web_Interface.git
cd RPi_Cam_Web_Interface/
./install.sh

In case of error

A message like:

#mmal: mmal_component_create_core: could not create component 'vc.ril.camera' (1)
sh: 1: /var/www/html/macros/error_hard.sh: Permission denied

Indicates that you might have a cabling problem.

In case of success

http://<yourpi>/html should show the cam interface:

Servo Interface - Servoblaster

• https://github.com/richardghirst/PiBits/tree/master/ServoBlaster
• https://github.com/srcshelton/servoblaster

wf@pibee:~ $ cd /usr/local/src
wf@pibee:/usr/local/src $ sudo git clone https://github.com/srcshelton/servoblaster
cd servoblaster
sudo make install
sudo reboot

OpenJDK

sudo apt-get install openjdk-8-jdk

Screen

sudo apt-get install screen

[edit]

Tested on

1. Mac OS 10.13.6 High Sierra with macports 2.5.4
2. Ubuntu 18.04.2 LTS

OpenCV

see also OpenCV

MacOS

If you'd like to generally use OpenCV you can get it e.g. from the macports project:

sudo port install opencv +java
ls /opt/local/share/OpenCV/java
libopencv_java343.dylib opencv-343.jar

For your convenience you'll find these two files in the lib directory.

Ubuntu 18

You might want to use scripts/opencvubuntu to download the 75 MByte libopencv_java343.so and install the backports libjasper libraries that are not published with Ubuntu 18 by default. You'll find some other links in the script that might also give you hints on how to proceed if you have another environment or different needs. You might want to create an Issue if you run into trouble.

Windows

Help Wanted -please file an issue if you want to things running on Windows.

[edit]

Deploying to your Raspberry is done by compiling e.g. on a laptop and then transferring the results to your PI.

Compile

scripts/install.sh

There is a usage for this script if you call it with scripts/install.sh -h. Essentially this is doing a mvn clean install. The clean part is important to get the platform specific native openCV Library integrated into the server code.

Example Configuration

You need to configure your PI's host address and user in a file ~/dukes/dukes.ini in the remotecar section. Also you need to check your GPIO pins and the servo settings for your car for LED,Engine and Wheels. You might want to try these things out before using the software the first time.

The original wiring is:

• MOTOR: GPIO-17 - Servoblaster ID 1
• WHEEL: GPIO-18 - Servoblaster ID 2
• LED: GPIO-24 - Servolbaster ID 6

The software uses the standard Servoblaster ID's above which are hard-coded at this time. The GPIO configuration will only work with other devices which we intend to support in the future.

#
# WF 2019-06-21
#
# rc-duke configuration file
# remote car raspberry PI configuration
remotecar.host=2.0.0.111
remotecar.user=wf
# camera url format
camera.url=http://2.0.0.111/html/cam_pic_new.php
# webserver configuration
webcontrol.port=8080
imageview.port=8081
# watchdog
# how often to send a heart beat in milliseconds
watchdog.heartbeat.interval.ms=150
# how many beats may be missed before remote car is forced to stop
watchdog.max.missed.beats=6
# be careful with your GPIO configuration
# you might damage your PI with a misconfiguration!
# led configuration
led.gpio=24
led.on=250
led.off=0
# wheel configuration
wheel.gpio=18
wheel.center=130
wheel.stepsize=5
wheel.maxleft=100
wheel.maxright=160
# engine configuration
engine.gpio=17
engine.speed.zero=130
engine.stepsize=1
engine.min.speed.reverse=130
engine.max.speed.reverse=130
engine.min.speed.forward=155
engine.max.speed.forward=157

The deploy script as a help option that shows how it can be used:

deploy.sh [-a|-d|-h|-m|-s]

  -a |--autostart         : configure the remotecar app to autostart on reboot
  -d |--debug             : debug this script
  -m |--maven             : run maven install
  -h |--help              : show this usage
  -s |--start             : deploy and start remotecar (duke) fat jar

• deploy -m will create the duke-farm fat jar
• deploy -s will send the jar to your PI using rsync and start it
• deploy -a will create an autostart configuration that restarts the client on reboot see https://github.com/rc-dukes/dukes/issues/26

Running the server

- start 'CarServer' in module 'rc-server' in IDE - point your browser to http://localhost:8080

Creating a fatjar of the server

../scripts/install.sh -h
usage: install.sh [-d|-f|-j|-q]* [-h]?

  -d |--debug        : debug this script
  -f |--fatjar       : create a fat jar
  -j |--javadoc      : with javadoc (default is without)
  -q |--quick        : no tests, no javadoc

  -h |--help         : show this usage

From the rc-server directory ../scripts/install.sh -f -q should create a fat jar with the maven assembly-plugin

java -jar target/rc-server-0.0.2-jar-with-dependencies.jar

(Please replace 0.0.2 with the current version of the project)

Should then start the server and the web ui should be available by either using

http://localhost:8080 or using the index.html file or detect.html files from the sources in rc-webcontrol/src/main/resources/web/

Starting all together

The start script will let you pick the components you'd like to start.

scripts/start -h
start [-d|-h]

  -a |--all     : run all parts of the software
  -c |--car     : run car software on pi
  -s |--server  : run server software
  -r |--remote  : point webbrowser to remot control
  -w |--web     : point webbrowser to detect.html
  -d |--debug   : debug this script
  -h |--help    : show this usage

So scripts/start -a will start

1. the car software on the pi
2. the server software e.g. on you laptop
3. point your webrowser to the detect.html page (after waiting for the page to get available)

[edit]

These are just some experiments with the Raspberry PI, Ultrasonic Sensors and Servos that might you get warmed up for the real challenge of getting your 1:10 RC self driving car going. You don't need to go thru all theses experiments. The basic car only uses the camera and two servos and no ultrasonic sensors.

1. Ultrasonic Sensor
2. SG90-Servo
3. Raspberry PI Camera
4. Raspberry PWM LED Control

Author: Wolfgang Fahl